Machine-learning tool may help predict acute kidney injury in patients with COVID-19 – Healio

Last Updated on November 17, 2020 by

November 16, 2020

1 min read



Chaudhary K, et al. Paper SA-OR02. Presented at: ASN Kidney Week. Oct. 22-25, 2020 (virtual meeting).

Healio Nephrology was unable to determine relevant financial disclosures prior to publication.

We were unable to process your request. Please try again later. If you continue to have this issue please contact

Researchers from Icahn School of Medicine at Mount Sinai found a machine-learning model performed well in predicting risk for AKI and need for dialysis in patients hospitalized with COVID-19.

Study results were presented at the virtual ASN Kidney Week.

AKI requiring dialysis
Source: Adobe Stock

“Preliminary reports indicate that acute kidney injury (AKI) is common in coronavirus disease (COVID-19) patients and is associated with worse outcomes,” Kumardeep Chaudhary, PhD, and colleagues wrote in a poster. “Identification of patients at high risk for developing severe AKI in hospitalized COVID-19 patients in the United States is not well-described.”

For the study, Chaudhary and colleagues included 3,235 patients who were admitted to the Mount Sinai Health System between Feb. 27 and April 15, 2020. The machine-learning model — which incorporated demographics, laboratory values and vital signs of patients within 48 hours of admission — was compared with fivefold cross validation to predict AKI requiring dialysis.

Overall, 46% of patients developed AKI and 20% developed the condition with the need for dialysis.

Results indicated the model had high accuracy with an area under the curve of 0.79.

Serum creatinine, age, potassium and heart rate were found to have the most impact.

Lili Chan

In a related press release, study co-author Lili Chan, MD, MS, commented on the utility of the findings.

“A machine-learning model using admission features had good performance for prediction of dialysis need,” she said. “Models like this are potentially useful for resource allocation and planning during future COVID-19 surges. We are in the process of deploying this model into our health care systems to help clinicians better care for their patients.”

Leave a Comment


Table of Contents